Friday, December 26, 2008

Analyse the Descriptive Data

Introduction

The type of analysis that is appropriate for your data depends on: your question, your study design and the type of data you collect.

Whatever your question, descriptive analysis will probably help you and your reader to understand your study better.

Remember that descriptive studies, correlation studies, and experimental & quasi-experimental studies will all use descriptive analysis

There are 3 types of data: nominal, ordinal and metric

Nominal are categories of data: For example "yes" and "no". They are names of groups. No group is higher or lower than others, they are just different.

Ordinal data categories of data that have some order. For example, "low", "medium" & "high", or the common Likert Scale "1 highly agree", "2 agree", "3 indifferent", "4 disagree" & "5 highly disagree". They are able to be ranked in some order, but the difference between each group may not be equal.

Metric data has equal intervals between two numbers. Fro example, the difference between 3 degrees and 4 degrees is the same as the difference between 19 degrees and 20 degrees Celsius. Metric data can be interval or ration scale data. Interval scale has an equal interval but an arbitrary zero (celsius), ratio scale has a real zero (time, length)

Descriptive analysis can be performed on all types of data. Descriptive analysis of your data will therefore be used in nearly every study.

Use of Descriptive Data Analysis

Descriptive data analysis can be used to summarise the data, provide evidence for validity or realiability of quantative data, and provide evidence for theories.

Descriptive Analysis of Nominal Data

Nominal data is typically collected through interview, questionnaires and observation, either as a case study or as a survey.

Data from these sources are usually coded to simplify their descriptions. This coding can be arranged prior to data collection (by preparing closed-ended questions and observation checklists) or the data can be collected without restriction (using open-ended questions and notes on observations and so know range)

Statistics can then be used to describe the "middle", the "spread" and also the frequencies and percentages in each category.

The mode, the most frequently occurring category, is the measure of "central tendency" to use with nominal data.

The range, can be used to describe the spread or variability of the data. However, remember the order of the nominal data is unimportant so saying the range was from group 1 to group 5 has no more meaning tahn saying the range was from group 1 to group 2. It is therefore often best to describe the spread of the data verbally.

Information about the pattern of the data can be used summarized using frequencies and percentages of each category

Descriptive Analysis of Ordinal Data

Ordinal data is typically collected through interviews, questionnaire, observation and testing. As with nominal data, statistics can then be used to describe the "middle", the "spread" and also the frequencies and percentages in each category.

The median, the middle score when all teh scores are listed in order, is the measure of central tendency to sue with ordinal data.

The range, the lowest to highest scores, can be used to describe the spread or variability of the data. Information about the pattern of the data can be summarised using frequencies and percentages at each level.

Descriptive Analysis of Metric Data

Metric data is typically collected testing

Statistics can tehn be used to describe the "middle", the "spread" and also the frequencies and percentages in each category.

The mean, the sum can be used to describe the spread or variability of the data. However, the variance and standard deviation can also be used, with standard deviation the usual measure. Information about the pattern of the data can be summarised using frequencies and percentages at each level

No comments: